Web Mining
Based on tutorials and presentations:
J. Han, D. Jing, W. Yan, Z. Xuan, M. Morzy, M. Chen, M. Brobbey, N. Somasetty, N. Niu,
P. Sundaram, S. Sajja, S. Thota, H. Ahonen-Myka, R. Cooley, B. Mobasher, J. Srivastava,
Y. Even-Zohar, A. Rajaraman and others
Discovering Knowledge from and about WWW is one of the basic abilities of an intelligent agent
WWW
Knowledge
2
Contents
 Introduction
 Web content mining
 Web structure mining



Evaluation of Web pages
HITS algorithm
Discovering cyber-communities on the Web
 Web usage mining
 Search engines for Web mining
 Multi-Layered Meta Web
3
Introduction
Data Mining and Web Mining
 Data mining: turn data into knowledge.
 Web mining is to apply data mining
techniques to extract and uncover
knowledge from web documents and
services.
5
WWW Specifics
 Web: A huge, widely-distributed, highly
heterogeneous, semi-structured,
hypertext/hypermedia, interconnected
information repository
 Web is a huge collection of documents plus

Hyper-link information

Access and usage information
6
A Few Themes in Web Mining
 Some interesting problems on Web mining

Mining what Web search engine finds

Identification of authoritative Web pages

Identification of Web communities

Web document classification

Warehousing a Meta-Web: Web yellow page service

Weblog mining (usage, access, and evolution)

Intelligent query answering in Web search
7
Web Mining taxonomy
 Web Content Mining

Web Page Content Mining
 Web Structure Mining


Search Result Mining
Capturing Web’s structure using link
interconnections
 Web Usage Mining


General Access Pattern Mining
Customized Usage Tracking
8
Web Content Mining
What is text mining?
 Data mining in text: find something useful and
surprising from a text collection;
 text mining vs. information retrieval;
 data mining vs. database queries.
10
Types of text mining
 Keyword (or term) based association analysis
 automatic document (topic) classification
 similarity detection


cluster documents by a common author
cluster documents containing information from a
common source
 sequence analysis: predicting a recurring event,
discovering trends
 anomaly detection: find information that violates
usual patterns
11
Types of text mining (cont.)
 discovery of frequent phrases
 text segmentation (into logical chunks)
 event detection and tracking
12
Information Retrieval
 Given:


A source of textual
documents
A user query (text
based)
Documents
source
Query
IR
System
• Find:
•
Document
A set (ranked) of documents that
are relevant to the query
Ranked
Documents
Document
Document
13
Intelligent Information Retrieval
 meaning of words


Synonyms “buy” / “purchase”
Ambiguity “bat” (baseball vs. mammal)
 order of words in the query


hot dog stand in the amusement park
hot amusement stand in the dog park
 user dependency for the data


direct feedback
indirect feedback
 authority of the source

IBM is more likely to be an authorized source then my second
14
far cousin
Intelligent Web Search
 Combine the intelligent IR tools




meaning of words
order of words in the query
user dependency for the data
authority of the source
 With the unique web features


retrieve Hyper-link information
utilize Hyper-link as input
15
What is Information Extraction?
 Given:


A source of textual documents
A well defined limited query (text based)
 Find:



Sentences with relevant information
Extract the relevant information and
ignore non-relevant information (important!)
Link related information and output in a
predetermined format
16
Information Extraction: Example
 Salvadoran President-elect Alfredo Cristiania condemned the terrorist killing of
Attorney General Roberto Garcia Alvarado and accused the Farabundo Marti
Natinal Liberation Front (FMLN) of the crime. … Garcia Alvarado, 56, was
killed when a bomb placed by urban guerillas on his vehicle exploded as it came
to a halt at an intersection in downtown San Salvador. … According to the
police and Garcia Alvarado’s driver, who escaped unscathed, the attorney
general was traveling with two bodyguards. One of them was injured.




Incident Date: 19 Apr 89
Incident Type: Bombing
Perpetrator Individual ID: “urban guerillas”
Human Target Name: “Roberto Garcia Alvarado”
 ...
17
Querying Extracted Information
Documents
source
Query 1
(E.g. job title)
Query 2
Extraction
System
(E.g. salary)
Combine
Query Results
Relevant Info 1
Ranked
Documents
Relevant Info 2
Relevant Info 3
18
What is Clustering ?
 Given:


A source of textual
documents
Similarity measure
Documents
source
Similarity
measure
Clustering
System
• e.g., how many words
are common in these
documents
• Find:
•
Several clusters of documents
that are relevant to each other
Doc
Doc
Doc
Doc
Doc
Doc
Do
Doc
Docc
Doc
19
Text Classification definition
 Given: a collection of labeled records (training set)
 Each record contains a set of features (attributes), and
the true class (label)
 Find: a model for the class as a function of the values of
the features
 Goal: previously unseen records should be assigned a
class as accurately as possible
 A test set is used to determine the accuracy of the
model. Usually, the given data set is divided into
training and test sets, with training set used to build the
model and test set used to validate it
20
Text Classification: An Example
Ex#
Hooligan
1
2
3
4
5
6
7
8
An English football fan
…
During a game in Italy
…
England has been
beating France …
Italian football fans were
cheering …
An average USA
salesman earns 75K
The game in London
was horrific
Manchester city is likely
to win the championship
Rome is taking the lead
in the football league
Yes
Hooligan
Yes
Yes
No
A Danish football fan
?
Turkey is playing vs. France.
The Turkish fans …
?
10
No
Yes
Test
Set
Yes
Yes
10
Training
Set
Learn
Classifier
Model
21
Discovery of frequent sequences (1)
 Find all frequent maximal sequences of words
(=phrases) from a collection of documents



frequent: frequency threshold is given; e.g. a phrase
has to occur in at least 15 documents
maximal: a phrase is not included in another longer
frequent phrase
other words are allowed between the words of a
sequence in text
22
Discovery of frequent sequences (2)
 Frequency of a sequence cannot be decided
locally: all the instances in the collection has to be
counted
 however: already a document of length 20
contains over million sequences
 only small fraction of sequences are frequent
23
Basic idea: bottom-up
 1. Collect all pairs from the documents, count
them, and select the frequent ones
 2. Build sequences of length p + 1 from frequent
sequences of length p
 3. Select sequences that are frequent
 4. Select maximal sequences
24
Summary
 There are many scientific and statistical text mining methods
developed, see e.g.:

http://www.cs.utexas.edu/users/pebronia/text-mining/

http://filebox.vt.edu/users/wfan/text_mining.html
 Also, it is important to study theoretical foundations of data
mining.

Data Mining Concepts and Techniques / J.Han & M.Kamber

Machine Learning, / T.Mitchell
25
Web Structure Mining
Web Structure Mining
 (1970) Researchers proposed methods of using
citations among journal articles to evaluate the quality
of reserach papers.
 Customer behavior – evaluate a quality of a product
based on the opinions of other customers (instead of
product’s description or advertisement)
 Unlike journal citations, the Web linkage has some
unique features:



not every hiperlink represents the endorsement we seek
one authority page will seldom have its Web page point to its
competitive authorities (CocaCola  Pepsi)
authoritative pages are seldom descriptive (Yahoo! may not
27
contain the description „Web search engine”)
Evaluation of Web pages
Web Search
 There are two approches:


page rank: for discovering the most important
pages on the Web (as used in Google)
hubs and authorities: a more detailed evaluation
of the importance of Web pages
 Basic definition of importance:

A page is important if important pages link to it
29
Predecessors and Successors of a Web
Page
Predecessors
Successors
…
…
30
Page Rank (1)
Simple solution: create a stochastic
matrix of the Web:
– Each page i corresponds to row i and
column i of the matrix
– If page j has n successors (links) then
the ijth cell of the matrix is equal to 1/n if
page i is one of these n succesors of
page j, and 0 otherwise.
31
Page Rank (2)
The intuition behind this matrix:
 initially each page has 1 unit of importance. At each
round, each page shares importance it has among its
successors, and receives new importance from its
predecessors.
 The importance of each page reaches a limit after
some steps
 That importance is also the probability that a Web
surfer, starting at a random page, and following
random links from each page will be at the page in
question after a long series of links.
32
Page Rank (3) – Example 1
 Assume that the Web consists of only three pages - A, B,
and C. The links among these pages are shown below.
Let [a, b, c] be
the vector of
importances for
these three pages
A
C
B
A
B
C
A
1/2
1/2
0
B
1/2
0
1
C
0
1/2
0
33
Page Rank – Example 1 (cont.)
 The equation describing the asymptotic values of these
three variables is:
a
1/2
1/2
0
a
b =
1/2
0
1
b
c
0
1/2
0
c
We can solve the equations like this one by starting with the
assumption a = b = c = 1, and applying the matrix to the current
estimate of these values repeatedly. The first four iterations give the
following estimates:
a=
b=
c=
1
1
1
1
3/2
1/2
5/4
1
3/4
9/8
11/8
1/2
5/4
…
17/16 …
11/16 ...
6/5
6/5
3/5
34
Problems with Real Web Graphs
 In the limit, the solution is a=b=6/5, c=3/5.
That is, a and b each have the same
importance, and twice of c.
 Problems with Real Web Graphs


dead ends: a page that has no succesors has nowhere to
send its importance.
spider traps: a group of one or more pages that have no
links out.
35
Page Rank – Example 2
 Assume now that the structure of the Web has changed.
The new matrix describing transitions is:
A
C
B
a
½
b = ½
c
0
½
0
1
0
0
0
a
b
c
The first four steps of the iterative solution
are:
a = 1 1 3/4 5/8 1/2
b = 1 1/2 1/2 3/8 5/16
c = 1 1/2 1/4 1/4 3/16
Eventually, each of a, b, and c become 0.
36
Page Rank – Example 3
• Assume now once more that the structure of the Web
has changed. The new matrix describing transitions is:
A
C
B
a
½
b = ½
c
0
½
0
1
0
0
1/2
a
b
c
The first four steps of the iterative solution
are:
a=1 1
3/4 5/8 1/2
b = 1 1/2 1/2 3/8 5/16
c = 1 3/2 7/4 2
35/16
c converges to 3, and a=b=0.
37
Google Solution
 Instead of applying the matrix directly, „tax” each page
some fraction of its current importance, and distribute
the taxed importance equally among all pages.
 Example: if we use 20% tax, the equation of the
previous example becomes:
a = 0.8 * (½*a + ½ *b +0*c)
b = 0.8 * (½*a + 0*b + 0*c)
c = 0.8 * (0*a + ½*b + 1*c)
The solution to this equation is a=7/11, b=5/11, and
c=21/11
38
Google Anti-Spam Solution
 „Spamming” is the attept by many Web sites to appear to
be about a subject that will attract surfers, without truly
being about that subject.
 Solutions:


Google tries to match words in your query to the words on the Web pages.
Unlike other search engines, Google tends to belive what others say about
you in their anchor text, making it harder from you to appear to be about
something you are not.
The use of Page Rank to measure importance also protects against
spammers. The naive measure (number of links into the page) can easily be
fooled by the spammers who creates 1000 pages that mutually link to one
another, while Page Rank recognizes that none of the pages have any real
importance.
39
PageRank Calculation
40
HITS Algorithm
--Topic Distillation on WWW
 Proposed by Jon M. Kleinberg
 Hyperlink-Induced Topic Search
41
Key Definitions
 Authorities
Relevant pages of the highest quality
on a broad topic
 Hubs
Pages that link to a collection of
authoritative pages on a broad topic
42
Hub-Authority Relations
Hubs
Authorities
43
Hyperlink-Induced Topic Search (HITS)
The approach consists of two phases:



It uses the query terms to collect a starting set of pages (200
pages) from an index-based search engine – root set of pages.
The root set is expanded into a base set by including all the
pages that the root set pages link to, and all the pages that link
to a page in the root set, up to a designed size cutoff, such as
2000-5000.
A weight-propagation phase is initiated. This is an iterative
process that determines numerical estimates of hub and
authority weights
44
Hub and Authorities




Define a matrix A whose rows and columns correspond to
Web pages with entry Aij=1 if page i links to page j, and 0
if not.
Let a and h be vectors, whose ith component corresponds to
the degrees of authority and hubbiness of the ith page.
Then:
h = A × a. That is, the hubbiness of each page is the sum
of the authorities of all the pages it links to.
a = AT × h. That is, the authority of each page is the sum of
the hubbiness of all the pages that link to it (AT - transponed
matrix).
Then, a = AT × A × a
h = A × AT × h
45
Hub and Authorities - Example
Consider the Web presented below.
A=
A
1 1 1
0 0 1
1 1 0
1
AT = 1
1
0
0
1
B
1
1
0
3 1 2
AAT = 1 1 0
2 0 2
C
ATA =
2 2 1
2 2 1
1 1 2
46
Hub and Authorities - Example
If we assume that the vectors
h = [ ha, hb, hc ] and a = [ aa, ab, ac ]
are each initially [ 1,1,1 ], the first three iterations
of the equations for a and h are the following:
aa = 1
5
24
114
ab = 1
5
24
114
ac = 1
4
18
84
ha = 1
6
28
132
hb = 1
2
8
36
hc = 1
4
20
96
47
Discovering cyber-communities
on the web
Based on link structure
What is cyber-community
 Defn: a community on the web is a group of
web pages sharing a common interest


Eg. A group of web pages talking about POP Music
Eg. A group of web pages interested in data-mining
 Main properties:



Pages in the same community should be similar to
each other in contents
The pages in one community should differ from the
pages in another community
Similar to cluster
49
Recursive Web Communities
 Definition: A community consists of members that
have more links within the community than
outside of the community.
 Community identification is NP-complete task
50
Two different types of communities
 Explicitly-defined
communities

They are well known ones,
such as the resource listed by
Yahoo!
 Implicitly-defined
communities

They are communities
unexpected or invisible to
most users
eg.
Arts
Music
Classic
Painting
Pop
eg. The group of web
pages interested in a
particular singer
51
Similarity of web pages
 Discovering web communities is similar to
clustering. For clustering, we must define the
similarity of two nodes
 A Method I:

For page and page B, A is related to B if there is a
hyper-link from A to B, or from B to A
Page A
Page B

Not so good. Consider the home page of IBM and
Microsoft.
52
Similarity of web pages
 Method II (from Bibliometrics)

Co-citation: the similarity of A and B is measured by the
number of pages cite both A and B
Page A
Page B
The normalized degree of
overlap in inbound links

Page A
Bibliographic coupling: the similarity of A and B is
measured by the number of pages cited by both A and B.
Page B
The normalized degree of
overlap in outbound links
53
Simple Cases (co-citations and coupling)
Page C
Page A
Page B
Better not to account self-citations
Page A
Page B
Number of pages for similarity
decision should be big enough
54
Example method of clustering
 The method from R. Kumar, P. Raghavan, S.
Rajagopalan, Andrew Tomkins

IBM Almaden Research Center
 They call their method communities trawling (CT)
 They implemented it on the graph of 200 millions
pages, it worked very well
55
Basic idea of CT
 Bipartite graph: Nodes
are partitioned into two
sets, F and C
 Every directed edge in
the graph is directed
from a node in F to a
node in C
F
C
F
C
56
Basic idea of CT
 Definition Bipartite cores



a complete bipartite
subgraph with at least i
nodes from F and at least j
nodes from C
i and j are tunable
parameters
A (i, j) Bipartite core
A (i=3, j=3) bipartite core
 Every community have such a
core with a certain i and j
57
Basic idea of CT
 A bipartite core is the identity of a community
 To extract all the communities is to enumerate all
the bipartite cores on the web
58
Web Communities
59
Read More
 http://webselforganization.com/
60
Web Usage Mining
What is Web Usage Mining?
 A Web is a collection of inter-related files on one or
more Web servers.
 Web Usage Mining.

Discovery of meaningful patterns from data generated by client-server
transactions.
 Typical Sources of Data:



automatically generated data stored in server access logs, referrer logs,
agent logs, and client-side cookies.
user profiles.
metadata: page attributes, content attributes, usage data.
62
Web Usage Mining (WUM)
The discovery of interesting user access patterns from Web
server logs
Generate simple statistical reports:
A summary report of hits and bytes transferred
A list of top requested URLs
A list of top referrers
A list of most common browsers used
Hits per hour/day/week/month reports
Hits per domain reports
Learn:
Who is visiting you site
The path visitors take through your pages
How much time visitors spend on each page
The most common starting page
Where visitors are leaving your site
63
Web Usage Mining – Three Phases
Pre-Processing
Raw
Sever log
Pattern Discovery
User session
File
Pattern Analysis
Rules and Patterns
Interesting
Knowledge
64
The Web Usage Mining Process
- General Architecture for the WEBMINER -
65
Web Server Access Logs
 Typical Data in a Server Access Log
looney.cs.umn.edu han - [09/Aug/1996:09:53:52 -0500] "GET mobasher/courses/cs5106/cs5106l1.html HTTP/1.0" 200
mega.cs.umn.edu njain - [09/Aug/1996:09:53:52 -0500] "GET / HTTP/1.0" 200 3291
mega.cs.umn.edu njain - [09/Aug/1996:09:53:53 -0500] "GET /images/backgnds/paper.gif HTTP/1.0" 200 3014
mega.cs.umn.edu njain - [09/Aug/1996:09:54:12 -0500] "GET /cgi-bin/Count.cgi?df=CS home.dat\&dd=C\&ft=1 HTTP
mega.cs.umn.edu njain - [09/Aug/1996:09:54:18 -0500] "GET advisor HTTP/1.0" 302
mega.cs.umn.edu njain - [09/Aug/1996:09:54:19 -0500] "GET advisor/ HTTP/1.0" 200 487
looney.cs.umn.edu han - [09/Aug/1996:09:54:28 -0500] "GET mobasher/courses/cs5106/cs5106l2.html HTTP/1.0" 200
...

...
...
Access Log Format
IP address userid time method url protocol status size
66
Example: Session Inference with Referrer Log
IP
Time
URL
Agent
Referrer
1
2
3
4
www.aol.com
www.aol.com
www.aol.com
www.aol.com
08:30:00
08:30:01
08:30:02
08:30:01
A
B
C
B
#
E
B
#
Mozillar/2.0;
Mozillar/2.0;
Mozillar/2.0;
Mozillar/2.0;
5
www.aol.com
08:30:03
C
B
Mozillar/2.0; Win 95
6
7
www.aol.com
www.aol.com
08:30:04
08:30:04
F
B
#
A
Mozillar/2.0; Win 95
Mozillar/2.0; AIX 4.1.4
8
www.aol.com
08:30:05
G
B
Mozillar/2.0; AIX 4.1.4
Identified Sessions:
S1:
# ==> A ==> B ==> G
S2:
E ==> B ==> C
S3:
# ==> B ==> C
S4:
# ==> F
from
from
from
from
AIX
AIX
AIX
Win
4.1.4
4.1.4
4.1.4
95
references 1, 7, 8
references 2, 3
references 4, 5
reference 6
67
Data Mining on Web Transactions
 Association Rules:

discovers similarity among sets of items across transactions

X =====> Y
where X, Y are sets of items, confidence or P(X v Y),
support or P(X^Y)
 Examples:



60% of clients who accessed /products/, also accessed
/products/software/webminer.htm.
30% of clients who accessed /special-offer.html, placed an online
order in /products/software/.
(Actual Example from IBM official Olympics Site)
{Badminton, Diving} ===> {Table Tennis} (69.7%,.35%)
68
Other Data Mining Techniques
 Sequential Patterns:


30% of clients who visited /products/software/, had done a search
in Yahoo using the keyword “software” before their visit
60% of clients who placed an online order for WEBMINER, placed
another online order for software within 15 days
 Clustering and Classification



clients who often access /products/software/webminer.html
tend to be from educational institutions.
clients who placed an online order for software tend to be students in the
20-25 age group and live in the United States.
75% of clients who download software from /products/software/demos/
visit between 7:00 and 11:00 pm on weekends.
69
Path and Usage Pattern Discovery
 Types of Path/Usage Information



Most Frequent paths traversed by users
Entry and Exit Points
Distribution of user session duration
 Examples:



60% of clients who accessed
/home/products/file1.html, followed the path
/home ==> /home/whatsnew ==> /home/products
==> /home/products/file1.html
(Olympics Web site) 30% of clients who accessed sport
specific pages started from the Sneakpeek page.
65% of clients left the site after 4 or less references.
70
Search Engines for Web
Mining
71
The number of Internet hosts exceeded...






1.000 in 1984
10.000 in 1987
100.000 in 1989
1.000.000 in 1992
10.000.000 in 1996
100.000.000 in 2000
72
Web search basics
Sponsored Links
CG Appliance Express
Discount Appliances (650) 756-3931
Same Day Certified Installation
www.cgappliance.com
San Francisco-Oakland-San Jose,
CA
User
Miele Vacuum Cleaners
Miele Vacuums- Complete Selection
Free Shipping!
www.vacuums.com
Miele Vacuum Cleaners
Miele-Free Air shipping!
All models. Helpful advice.
www.best-vacuum.com
Web
Results 1 - 10 of about 7,310,000 for miele. (0.12 seconds)
Miele, Inc -- Anything else is a compromise
Web crawler
At the heart of your home, Appliances by Miele. ... USA. to miele.com. Residential Appliances.
Vacuum Cleaners. Dishwashers. Cooking Appliances. Steam Oven. Coffee System ...
www.miele.com/ - 20k - Cached - Similar pages
Miele
Welcome to Miele, the home of the very best appliances and kitchens in the world.
www.miele.co.uk/ - 3k - Cached - Similar pages
Miele - Deutscher Hersteller von Einbaugeräten, Hausgeräten ... - [ Translate this
page ]
Das Portal zum Thema Essen & Geniessen online unter www.zu-tisch.de. Miele weltweit
...ein Leben lang. ... Wählen Sie die Miele Vertretung Ihres Landes.
www.miele.de/ - 10k - Cached - Similar pages
Herzlich willkommen bei Miele Österreich - [ Translate this page ]
Herzlich willkommen bei Miele Österreich Wenn Sie nicht automatisch
weitergeleitet werden, klicken Sie bitte hier! HAUSHALTSGERÄTE ...
www.miele.at/ - 3k - Cached - Similar pages
Search
Indexer
The Web
Indexes
73
Ad indexes
Search engine components
 Spider (a.k.a. crawler/robot) – builds corpus

Collects web pages recursively
• For each known URL, fetch the page, parse it, and extract new URLs
• Repeat

Additional pages from direct submissions & other sources
 The indexer – creates inverted indexes

Various policies wrt which words are indexed, capitalization,
support for Unicode, stemming, support for phrases, etc.
 Query processor – serves query results


Front end – query reformulation, word stemming,
capitalization, optimization of Booleans, etc.
Back end – finds matching documents and ranks them
74
Web Search Products and Services










Alta Vista
DB2 text extender
Excite
Fulcrum
Glimpse (Academic)
Google!
Inforseek Internet
Inforseek Intranet
Inktomi (HotBot)
Lycos





PLS
Smart (Academic)
Oracle text extender
Verity
Yahoo!
75
Boolean search in AltaVista
76
Specifying field content in HotBot
77
Natural language interface in AskJeeves
78
Three examples of search strategies
 Rank web pages based on popularity
 Rank web pages based on word frequency
 Match query to an expert database
All the major search engines use a mixed
strategy in ranking web pages and
responding to queries
79
Rank based on word frequency
 Library analogue: Keyword search
 Basic factors in HotBot ranking of pages:




words in the title
keyword meta tags
word frequency in the document
document length
80
Alternative word frequency measures
 Excite uses a thesaurus to search for what
you want, rather than what you ask for
 AltaVista allows you to look for words that
occur within a set distance of each other
 NorthernLight weighs results by search
term sequence, from left to right
81
Rank based on popularity
 Library analogue: citation index
 The Google strategy for ranking pages:



Rank is based on the number of links to a page
Pages with a high rank have a lot of other web
pages that link to it
The formula is on the Google help page 
82
More on popularity ranking
 The Google philosophy is also applied by
others, such as NorthernLight
 HotBot measures the popularity of a page
by how frequently users have clicked on it
in past search results
83
Expert databases: Yahoo!
 An expert database contains predefined
responses to common queries
 A simple approach is subject directory, e.g.
in Yahoo!, which contains a selection of
links for each topic
 The selection is small, but can be useful
84
Expert databases: AskJeeves
 AskJeeves has predefined responses to
various types of common queries
 These prepared answers are augmented by a
meta-search, which searches other SEs
 Library analogue: Reference desk
85
Best wines in France: AskJeeves
86
Best wines in France: HotBot
87
Best wines in France: Google
88
Some possible improvements
 Automatic translation of websites
 More natural language intelligence
 Use meta data on trusty web pages
89
Predicting the future...
 Association analysis of related documents
(a popular data mining technique)
 Graphical display of web communities
(both two- and three dimensional)
 Client-adjusted query responses
90
Multi-Layered Meta-Web
91
What Role will XML Play?
 XML provides a promising direction for a more structured
Web and DBMS-based Web servers
 Promote standardization, help construction of multi-layered
Web-base.
 Will XML transform the Web into one unified database
enabling structured queries like:

“find the cheapest airline ticket from NY to Chicago”

“list all jobs with salary > 50 K in the Boston area”
 It is a dream now but more will be minable in the future!
92
Web Mining in an XML View
 Suppose most of the documents on web will be
published in XML format and come with a valid
DTD.
 XML documents can be stored in a relational
database, OO database, or a specially-designed
database
 To increase efficiency, XML documents can be
stored in an intermediate format.
93
Mine What Web Search
Engine Finds
 Current Web search engines: convenient source for mining

keyword-based, return too many answers, low quality
answers, still missing a lot, not customized, etc.
 Data mining will help:

coverage: using synonyms and conceptual hierarchies

better search primitives: user preferences/hints

linkage analysis: authoritative pages and clusters

Web-based languages: XML + WebSQL + WebML

customization: home page + Weblog + user profiles
94
Warehousing a Meta-Web:
An MLDB Approach
 Meta-Web: A structure which summarizes the contents, structure,
linkage, and access of the Web and which evolves with the Web
 Layer0: the Web itself
 Layer1: the lowest layer of the Meta-Web

an entry: a Web page summary, including class, time, URL, contents,
keywords, popularity, weight, links, etc.
 Layer2 and up: summary/classification/clustering in various ways
and distributed for various applications
 Meta-Web can be warehoused and incrementally updated
 Querying and mining can be performed on or assisted by metaWeb (a multi-layer digital library catalogue, yellow page).
95
A Multiple Layered Meta-Web Architecture
Layern
More Generalized Descriptions
...
Layer1
Generalized Descriptions
Layer0
96
Construction of Multi-Layer Meta-Web
 XML: facilitates structured and meta-information extraction
 Hidden Web: DB schema “extraction” + other meta info
 Automatic classification of Web documents:

based on Yahoo!, etc. as training set + keyword-based
correlation/classification analysis (AI assistance)
 Automatic ranking of important Web pages

authoritative site recognition and clustering Web pages
 Generalization-based multi-layer meta-Web construction

With the assistance of clustering and classification analysis
97
Use of Multi-Layer Meta Web
 Benefits of Multi-Layer Meta-Web:

Multi-dimensional Web info summary analysis

Approximate and intelligent query answering

Web high-level query answering (WebSQL, WebML)

Web content and structure mining

Observing the dynamics/evolution of the Web
 Is it realistic to construct such a meta-Web?

Benefits even if it is partially constructed

Benefits may justify the cost of tool development,
standardization and partial restructuring
98
Conclusion
 Web Mining fills the information gap
between web users and web designers
99
100
101